Articles

Lenvatinib versus sorafenib in first-line treatment of

patients with unresectable hepatocellular carcinoma:
a randomised phase 3 non-inferiority trial
Masatoshi Kudo, Richard S Finn, Shukui Qin, Kwang-Hyub Han, Kenji Ikeda, Fabio Piscaglia, Ari Baron*, Joong-Won Park*, Guohong Han*,
Jacek Jassem, Jean Frederic Blanc, Arndt Vogel, Dmitry Komov, T R Jeffry Evans, Carlos Lopez, Corina Dutcus, Matthew Guo, Kenichi Saito,
Silvija Kraljevic, Toshiyuki Tamai, Min Ren, Ann-Lii Cheng

Summary
Background In a phase 2 trial, lenvatinib, an inhibitor of VEGF receptors 1–3, FGF receptors 1–4, PDGF receptor α, Lancet 2018; 391: 1163–73
RET, and KIT, showed activity in hepatocellular carcinoma. We aimed to compare overall survival in patients treated Published Online
with lenvatinib versus sorafenib as a first-line treatment for unresectable hepatocellular carcinoma. February 9, 2018
http://dx.doi.org/10.1016/
S0140-6736(18)30207-1
Methods This was an open-label, phase 3, multicentre, non-inferiority trial that recruited patients with unresectable
See Comment page 1123
hepatocellular carcinoma, who had not received treatment for advanced disease, at 154 sites in 20 countries throughout
*Contributed equally
the Asia-Pacific, European, and North American regions. Patients were randomly assigned (1:1) via an interactive
Department of
voice–web response system—with region; macroscopic portal vein invasion, extrahepatic spread, or both; Eastern Gastroenterology and
Cooperative Oncology Group performance status; and bodyweight as stratification factors—to receive oral lenvatinib Hepatology, Kindai University
(12 mg/day for bodyweight ≥60 kg or 8 mg/day for bodyweight <60 kg) or sorafenib 400 mg twice-daily in 28-day Faculty of Medicine, Osaka,
cycles. The primary endpoint was overall survival, measured from the date of randomisation until the date of death Japan (Prof M Kudo MD); Geffen
School of Medicine at UCLA,
from any cause. The efficacy analysis followed the intention-to-treat principle, and only patients who received Santa Monica, CA, USA
treatment were included in the safety analysis. The non-inferiority margin was set at 1·08. The trial is registered with (R S Finn MD); Nanjing Bayi
ClinicalTrials.gov, number NCT01761266. Hospital, Nanjing, Jiangsu,
China (Prof S Qin MD);
Severance Hospital, Yonsei
Findings Between March 1, 2013 and July 30, 2015, 1492 patients were recruited. 954 eligible patients were randomly University, Seoul, South Korea
assigned to lenvatinib (n=478) or sorafenib (n=476). Median survival time for lenvatinib of 13·6 months (95% CI (Prof K-H Han MD); Toranomon
12·1–14·9) was non-inferior to sorafenib (12·3 months, 10·4–13·9; hazard ratio 0·92, 95% CI 0·79–1·06), meeting Hospital, Tokyo, Japan
criteria for non-inferiority. The most common any-grade adverse events were hypertension (201 [42%]), diarrhoea (K Ikeda MD); University of
Bologna, Bologna, Italy
(184 [39%]), decreased appetite (162 [34%]), and decreased weight (147 [31%]) for lenvatinib, and palmar-plantar (Prof F Piscaglia MD); California
erythrodysaesthesia (249 [52%]), diarrhoea (220 [46%]), hypertension (144 [30%]), and decreased appetite (127 [27%]) Pacific Medical Center,
for sorafenib. San Francisco, CA, USA
(A Baron MD); National Cancer
Center Korea, Goyang-si,
Interpretation Lenvatinib was non-inferior to sorafenib in overall survival in untreated advanced hepatocellular South Korea (Prof J-W Park MD);
carcinoma. The safety and tolerability profiles of lenvatinib were consistent with those previously observed. Xijing Hospital, Fourth Military
Medical University, Xi’an,
China (Prof G Han MD); Medical
Funding Eisai Inc.
University of Gdansk, Gdansk,
Poland (Prof J Jassem MD);
Introduction phase 3 trials (of sunitinib, brivanib, linifanib, and University of Bordeaux,
Hepatocellular carcinoma is the most common type of erlotinib plus sorafenib) that did not show non-inferiority6–8 Bordeaux, France
(Prof J F Blanc MD); Hannover
liver cancer, which is the third leading cause of cancer or superiority9 to sorafenib in terms of overall survival in
Medical School, Hannover,
deaths worldwide, causing nearly 745 000 deaths each year.1 first-line treatment of hepatocellular carcinoma. No Germany (Prof A Vogel MD);
The disease usually occurs in people with chronic liver approved first-line systemic treatments are available for N N Blokhin Cancer Research
disease, particularly cirrhosis, which limits the feasibility advanced unresectable hepatocellular carcinoma other Center, Moscow, Russia
(Prof D Komov MD); University
of surgical resection.2,3 Sorafenib, an oral multikinase than sorafenib. Only regorafenib and nivolumab are of Glasgow, Beatson West of
inhibitor, is the only systemic therapy proven to extend approved as second-line systemic treatments for patients Scotland Cancer Centre,
overall survival when used as a first-line treatment, who do not respond to sorafenib.10 Otherwise, best Glasgow, UK
showing a median improvement of 2·8 months compared supportive care or participation in clinical trials is (Prof T R J Evans MD); Marqués
de Valdecilla University
with placebo (10·7 months vs 7·9 months; hazard ratio recommended in the second-line setting by treat­ Hospital, Santander, Spain
[HR] 0·69; p<0·001), despite a low response rate of 2%.4 In ment guidelines.11 Therefore, because of the paucity of (C Lopez PhD); Eisai, Woodcliff
patients from the Asia-Pacific region taking sorafenib, the systemic treatment options for patients with advanced Lake, NJ, USA (C Dutcus MD,
median improvement in overall survival compared with hepatocellular carcinoma, a need exists to develop new M Guo PhD, K Saito MS,
T Tamai MS, M Ren PhD); Eisai,
placebo was 2·3 months (6·5 months vs 4·2 months; drugs for effective management of this disease. Hatfield, UK (S Kraljevic MD);
HR 0·68; p=0·014).5 Lenvatinib is an oral multikinase inhibitor that targets and National Taiwan
Drug development for hepatocellular carcinoma in the VEGF receptors 1–3, FGF receptors 1–4, PDGF receptor α, University Hospital, Taipei,
past 10 years has been marked by four failed global RET, and KIT.12–15 Lenvatinib monotherapy is approved for Taiwan (Prof A-L Cheng MD)

www.thelancet.com Vol 391 March 24, 2018 1163

 Articles

Correspondence to:
Prof Masatoshi Kudo,
Department of Gastroenterology
and Hepatology,
Kindai University Faculty of
Medicine, 337-2 Ohno-Higashi,
Osaka, Japan
m-kudo@med.kindai.ac.jp
Research in context
Evidence before this study
We searched PubMed from inception up to March 16, 2017 using
the search terms “phase 3” [Title/Abstract] OR “phase III” [Title/
Abstract] AND “hepatocellular carcinoma” [MeSH Terms].
The search was restricted to clinical trials in English language only
and yielded 65 reports. Of these publications, 21 described the use
of targeted drugs for treatment of hepatocellular carcinoma,
11 were studies of single-drug sorafenib treatment, and three
were studies of sorafenib in combination with another drug.
Five trials investigated targeted agents following treatment with
sorafenib and four trials investigated first-line treatment of
hepatocellular carcinoma with sorafenib as the comparator. None
of these four trials met their primary endpoints of non-inferiority
or superiority over sorafenib in terms of overall survival.
Added value of this study
To our knowledge, this is the first global phase 3 trial in 10 years
to meet its primary endpoint of non-inferiority in terms of
overall survival against sorafenib as a first-line treatment for
hepatocellular carcinoma. Furthermore, lenvatinib showed
statistically significant and clinically meaningful improvement
in terms of all secondary endpoints (progression-free survival,
time to progression, and objective response rate) with a
reasonable safety profile.
Implications of all the available evidence
The results of this study support lenvatinib as a first-line
treatment option for patients with unresectable hepatocellular
carcinoma.

treatment of radioiodine-refractory differentiated thyroid
cancer.16 Lenvatinib and everolimus are approved as a
combined treatment for advanced renal cell carcinoma
following one previous antiangiogenic therapy.17 In a
phase 2 study of patients with advanced hepatocellular
carcinoma, 12 mg lenvatinib once-daily showed clinical
activity and had an acceptable safety profile.18 Based
on dose adjustments depending on bodyweight and
See Online for appendix pharmacokinetic modelling data,19 a starting dose of
lenvatinib was adopted (12 mg for patients ≥60 kg and
8 mg for patients <60 kg once-daily) for further clinical
development in hepatocellular carcinoma. Given the
efficacy signal observed in this phase 2 study,18 we did a
phase 3 randomised, open-label, non-inferiority study to
compare the efficacy and safety of lenvatinib versus
sorafenib as a first-line treatment for unresectable
hepatocellular carcinoma.
Methods
Study design and participants
This multicentre, phase 3, randomised, open-label, non-
inferiority study was done at 154 sites in 20 countries
throughout the Asia-Pacific, European, and North
American regions (China, Hong Kong, Japan, South Korea,
Malaysia, Philippines, Singapore, Taiwan, Thailand,
Belgium, Canada, France, Germany, Israel, Italy, Poland,
Russia, Spain, UK, and USA).
Eligible patients had unresectable hepatocellular
carcinoma, with diagnoses confirmed histologically or
cytologically, or confirmed clinically in accordance with
American Association for the Study of Liver Diseases
criteria. Included patients also had one or more measurable
target lesions (lesions previously treated with radiotherapy
or locoregional therapy had to show radiographic evidence
of disease progression to be deemed target lesions) based
on modified Response Evaluation Criteria in Solid
Tumours (mRECIST),20 Barcelona Clinic Liver Cancer
stage B or C categorisation,21 Child-Pugh class A, and an
Eastern Cooperative Oncology Group performance status
score of 0 or 1. All eligible patients had controlled blood
pressure (≤150/90 mm Hg), adequate liver function
(albumin ≥2·8 g/dL, bilirubin ≤3·0 mg/dL, and aspartate
aminotransferase, alkaline phosphatase, and alanine
aminotransferase ≤5 times the upper limit of normal), and
adequate bone marrow (haemoglobin ≥8·5 g/dL, platelet
count ≥75  × 
10⁹ per L, and absolute neutrophil count
≥1·5 × 10⁹ per L), blood (international normalised ratio
≤2·3), renal, and pancreatic function (see appendix for a
full list of inclusion criteria). Patients with 50% or higher
liver occupation, obvious invasion of the bile duct, or
invasion at the main portal vein were excluded from the
study. Patients were also excluded if they had received
previous systemic therapy for hepatocellular carcinoma
(see appendix for a full list of exclusion criteria).
All patients provided written informed consent before
undergoing any study-specific procedures. All relevant
institutional review boards approved the study, which
was done in accordance with the Declaration of Helsinki
and local laws.
Randomisation and masking
Patients were randomly assigned in a 1:1 ratio to receive
either lenvatinib or sorafenib. Allocation of treatment
group was done with an interactive voice–web res­
ponse system, which also functioned as the allocation
concealment method, with region (Asia-Pacific [defined
as China, Hong Kong, Japan, South Korea, Malaysia,
Philippines, Singapore, Taiwan, and Thailand] or western
[defined as Belgium, UK, Spain, Germany, Italy, Poland,
France, USA, Canada, Israel, and Russia]), macroscopic
portal vein invasion, extrahepatic spread, or both (yes or
no), Eastern Coop­erative Oncology Group performance
status (0 or 1), and bodyweight (<60 kg or ≥60 kg) as
stratification factors. A randomisation block size of 2 was
used. The randomisation sequence was generated by an
inde­pendent statistician by the system vendor, and the
investigators obtained the randomisation assignments
from the system directly. Because the study was open

1164 www.thelancet.com Vol 391 March 24, 2018

 Articles

label, the treatments were not masked to the patients or
investigators.
Procedures
Patients received oral lenvatinib (Eisai Inc., Woodcliff
Lake, NJ, USA) 12 mg/day (for bodyweight ≥60 kg) or
8 mg/day (for bodyweight <60 kg) or sorafenib (Bayer,
Leverkusen, Germany) 400 mg twice-daily in 28-day
cycles. Dose interruptions followed by reductions for
lenvatinib-related toxicities (to 8 mg and 4 mg/day, or
4 mg every other day) were permitted. Modifications to
sorafenib doses were implemented according to
prescribing information in each region (all patients in
the sorafenib arm received a starting dose of 400 mg
orally twice-daily).
Local investigators evaluated tumours in each treatment
arm in accordance with mRECIST.20,22 The liver was
examined with CT or MRI by use of a triphasic scanning
technique. Tumour assessments were done every 8 weeks
(irrespective of dose interruptions) until radiological
disease progression. Patients who discontinued study
treatment without disease pro­ gression had tumour
assessments every 8 weeks or until disease progression or
the start of another anticancer treatment. Safety
assessments were done throughout the study. Quality-of-
life questionnaires were adminis­tered at baseline, on day 1
of each subsequent treatment cycle, and at the off-
treatment visit, which occurred within 30 days of the final
administration of study drug. Quality of life was assessed
with the European Organisation for Research and
Treatment of Cancer Quality of Life Questionnaire C30
(EORTC QLQ-C30)23,24 and the hepato­cellular carcinoma-
specific EORTC QLQ-HCC1825 health questionnaires.
The follow-up period began immediately after the
off-treatment visit and was planned to continue if the
patient was alive or until the sponsor terminated the study,
or the patient withdrew consent. Patients were planned to
be followed up for survival every 12 weeks, and all
anticancer treatments received were reported.
the safety analysis set (all patients who received at
least one dose of study treatment). Post-hoc exploratory
tumour assessments using mRECIST and RECIST
version 1.1 were done by masked central independent
imaging review.
A population pharmacokinetic analysis for lenvatinib
was done to derive individual pharmacokinetic para­
meters and lenvatinib exposure for this study. The
dataset used in the analysis included lenvatinib plasma
concentrations from 468 patients with hepato­ cellular
carcinoma in this study, and lenvatinib plasma
concentration pooled from 12 additional studies
(phase 1–3) in healthy people and patients with other
tumour types (eg, differentiated thyroid cancer).
Statistical analysis
The primary endpoint of overall survival was first tested
for non-inferiority, then for superiority. Using a non-
inferiority test by the 95% CI lower-limit method on log
HR for overall survival with assumed true HR of 0·80
and a non-inferiority margin of 1·08 (corresponding to
1492 patients assessed for eligibility
538 ineligible
480 did not meet eligibility criteria
7 adverse events
2 lost to follow-up
35 withdrew consent
14 other
954 randomly assigned
478 assigned to lenvatinib 476 assigned to sorafenib
2 did not meet eligibility criteria 1 chose not to receive sorafenib

Outcomes 476 received lenvatinib 475 received sorafenib

The primary endpoint was overall survival, measured from

the date of randomisation until the date of death from any 449 discontinued treatment 450 discontinued treatment
cause. Patients who were lost to follow-up were censored at 311 radiological progression 347 radiological progression
the last date they were known to be alive, and patients who 63 adverse event 43 adverse event
32 clinical progression 33 clinical progression
remained alive were censored at the time of data cutoff. 28 patient’s choice 14 patient’s choice
Secondary endpoints were progression-free survival, 3 lost to follow-up 1 lost to follow-up
9 withdrew consent 5 withdrew consent
time to progression, objective response rate, quality-of-life 3 other 7 other
measurements, and plasma pharmacokinetics lenvatinib
exposure parameters. All efficacy evaluations were based
on the full analysis set (all randomised patients). 27 treatment ongoing 25 treatment ongoing

Safety assessments included recording of vital signs,

haematological and biochemical laboratory testing,
urinalysis, and electrocardiography. Adverse events were
graded according to the National Cancer Institute
Common Terminology Criteria for Adverse Events
version 4.0.26 All safety evaluations were based on
478 included in intention-to-treat analysis 476 included in intention-to-treat analysis
Figure 1: Trial profile
At the time of data cutoff (Nov 13, 2016; for the required 700 death events), 701 deaths had occurred (351 in the
lenvatinib arm, 350 in the sorafenib arm).

www.thelancet.com Vol 391 March 24, 2018 1165

 Articles

60% retention of sorafenib effect vs placebo, and set 82% using a superiority test with assumed true HR of
based on previous phase 3 trials of sorafenib4,5), the 0·80. The overall false positive rate was set at 0·05
power of the study to declare non-inferiority was (two-sided). Non-inferiority was declared if the upper
approximately 97%. The power of the study to declare limit of the two-sided 95% CI for HR was less than 1·08.
superiority of lenvatinib to sorafenib was approximately The required number of events for the primary analysis
was 700 deaths, assuming 5% dropout. HR and 95% CI
Lenvatinib (n=478) Sorafenib (n=476) Total (n=954) were estimated from a Cox proportional hazard model
Age (years), median 63·0 (20–88) 62·0 (22–88) 62·0 (20–88) with treatment group as a factor, and with the analysis
(range) stratified according to the same factors applied for
Age group (years) randomisation for primary and subgroup analyses
<65 270 (56%) 283 (59%) 553 (58%) where appropriate. For the subgroup analysis, analyses
≥65 to <75 150 (31%) 126 (26%) 276 (30%) were done within each subgroup.
≥75 58 (12%) 67 (14%) 125 (13%) A fixed sequence procedure was used to control the
Sex overall type I error rate of analyses for both the primary
Male 405 (85%) 401 (84%) 806 (84%) and secondary efficacy endpoints at α=0·05 (two-sided).
Female 73 (15%) 75 (16%) 148 (16%) After non-inferiority was declared, secondary efficacy
Region endpoints were tested. Differences in progression-free
Western 157 (33%) 157 (33%) 314 (33%) survival and time to progression were evaluated using a
Asia-Pacific 321 (67%) 319 (67%) 640 (67%) stratified log-rank test with randomisation stratification
Race factors, with the associated HR and 95% CI. The same
White 135 (28%) 141 (30%) 276 (29%) method was used to evaluate differences in pro­
Asian 334 (70%) 326 (68%) 660 (69%) gression-free survival and time to progression in the
Other 9 (2%) 9 (2%) 18 (2%)
subgroup analyses. A difference in the objective response
Bodyweight (kg)
rate was evaluated using the Cochran-Mantel-Haenszel
<60 153 (32%) 146 (31%) 299 (31%)
χ² test with randomisation stratification factors as strata,
with associated odds ratio (OR) and 95% CI. To assess
≥60 325 (68%) 330 (69%) 655 (69%)
futility, two interim analyses (at 30% and 70% of the
Eastern Cooperative Oncology Group performance status
target number of events) were done using Bayesian
0 304 (64%) 301 (63%) 605 (63%)
predictive probability in a non-inferiority design.
1 174 (36%) 175 (37%) 349 (37%)
The efficacy analysis followed the intention-to-treat
Child-Pugh class
principle. Only patients who received treatment were
A 475 (99%) 471 (99%) 946 (99%)
included in the safety analysis.
B 3 (1%) 5 (1%) 8 (1%)
Programming and statistical analyses were done with
Macroscopic portal vein invasion
SAS version 9 or higher. The study was overseen by an
Yes 109 (23%) 90 (19%) 199 (21%)
independent data monitoring committee. The study is
No 369 (77%) 386 (81%) 755 (79%)
registered with ClinicalTrials.gov, number NCT01761266.
Extrahepatic spread
Yes 291 (61%) 295 (62%) 586 (61%) Role of the funding source
No 187 (39%) 181 (38%) 368 (39%) The study was funded by Eisai Inc, (Woodcliff Lake, NJ,
Macroscopic portal vein invasion, extrahepatic spread, or both USA) and designed in collaboration with the principal
Yes 329 (69%) 336 (71%) 665 (70%) investigators. The funder employed CD, MG, KS, SK, TT,
No 149 (31%) 140 (29%) 289 (30%) and MR, who played a significant part in study design,
Underlying cirrhosis based on masked independent imaging review data collection, data analysis, data interpretation, and
Yes 356 (74%) 364 (76%) 720 (75%) writing of the report. The corresponding author had full
No 122 (26%) 112 (24%) 234 (25%) access to all data in the study and had final responsibility
Barcelona Clinic Liver Cancer stage for the decision to submit for publication.
B (intermediate stage) 104 (22%) 92 (19%) 196 (21%)
C (advanced stage) 374 (78%) 384 (81%) 758 (79%) Results
Involved disease sites Between March 1, 2013, and July 30, 2015, 1492 patients
Liver 441 (92%) 430 (90%) 871 (91%) were recruited. 954 eligible patients from 20 countries
Lung 163 (34%) 144 (30%) 307 (32%) were randomly assigned to receive lenvatinib (n=478) or
Involved disease sites per patient* sorafenib (n=476, figure 1).
1 207 (43%) 207 (43%) 414 (43%) Patient baseline characteristics were similar be­
2 167 (35%) 183 (38%) 350 (37%) tween treatment groups, except for baseline hepatitis C
≥3 103 (22%) 86 (18%) 189 (20%) aetiology and α-fetoprotein concentrations (table 1). At
(Table 1 continues on next page) the time of data cutoff (Nov 13, 2016, at 701 deaths), the
median duration of follow-up was 27·7 months

1166 www.thelancet.com Vol 391 March 24, 2018

 Articles

(IQR 23·3–32·8) in the lenvatinib group and 27·2 months

Lenvatinib (n=478) Sorafenib (n=476) Total (n=954)
(22·6–31·3) in the sorafenib group.
(Continued from previous page)
Lenvatinib showed non-inferiority in terms of overall
Aetiology of chronic liver disease
survival compared with sorafenib (figure 2). Median
overall survival duration was 13·6 months (95% CI Hepatitis B 251 (53%) 228 (48%) 479 (50%)

12·1–14·9) for 478 patients in the lenvatinib group, Hepatitis C 91 (19%) 126 (26%) 217 (23%)

compared with 12·3 months (10·4–13·9) for 476 patients Alcohol 36 (8%) 21 (4%) 57 (6%)
in the sorafenib group (HR 0·92, 95% CI 0·79–1·06, Other 38 (8%) 32 (7%) 70 (7%)
figure 2, table 2; results from the per-protocol analysis Unknown 62 (13%) 69 (14%) 131 (14%)
set are shown in the appendix). Overall survival Baseline α-fetoprotein concentration (ng/mL)
superiority over sorafenib was not achieved. The effect Number of patients 471 (99%) 463 (97%) 934 (98%)
of lenvatinib and sorafenib on median overall survival Mean (SD) 17 507·5 (105 137·4) 16 678·5 (94 789·5) 17 096·5 (100 088·8)
was consistent across sub­ groups based on baseline Median (IQR) 133·1 (8·0−3730·6) 71·2 (5·2−1081·8) 89·0 (6·3−2120·2)
characteristics (figure 3). Although baseline α-fetoprotein Baseline α-fetoprotein concentration group (ng/mL)
concentration was not a prespecified stratum, patients <200 255 (53%) 286 (60%) 541 (57%)
with baseline α-fetoprotein concen­ trations less than ≥200 222 (46%) 187 (39%) 409 (43%)
200 ng/mL had longer overall survival than did those Missing 1 (<1%) 3 (1%) 4 (<1%)
with α-fetoprotein concentration of at least 200 ng/mL Concomitant systemic 163 (34%) 149 (31%) 312 (33%)
in both treat­ment groups (figure 3). More patients had antiviral therapy for
hepatitis B or C
baseline α-fetoprotein levels less than 200 ng/mL in the
Previous therapy
sorafenib arm compared with the lenvatinib arm
Previous anticancer 327 (68%) 344 (72%) 671 (70%)
(table 1). procedures
Lenvatinib showed a statistically significant improve­
Radiotherapy 49 (10%) 60 (13%) 109 (11%)
ment compared with sorafenib for all secondary effi­
cacy endpoints as determined by investigator tumour Data are mean (SD) or n (%) unless otherwise specified. *One patient had no baseline target lesion.
assessments based on mRECIST. Median progression- Table 1: Demographic and disease characteristics at baseline
free survival for lenvatinib was longer than that for
sorafenib (table 2, figure 4). Median time to progression
was 8·9 months (95% CI 7·4–9·2) for patients in the
lenvatinib group compared to 3·7 months (3·6–5·4) for 100 Median overall survival duration
(months; 95% CI)
patients in the sorafenib group (table 2, appendix). 90
Lenvatinib 13·6 (12·1–14·9)
Lenvatinib also showed a greater objective response Sorafenib 12·3 (10·4–13·9)
80
rate than did sorafenib (table 2, appendix). Improvements
in all secondary efficacy endpoints (progression-free 70 HR 0·92 (95% CI 0·79–1·06)

survival, time to progression, and objective response)

Overall survival (%)

60
with lenvatinib compared to sorafenib were consistent
50
across all predefined subgroups (figure 3, appendix).
Analysis for overall survival with predefined subgroups 40
supports the robustness of the non-inferiority result 30
(appendix). Masked independent imaging review con­
20
firmed progression-free survival and time to progression
based on investigator assessments according to mRECIST 10
(table 2, figure 4). Similar progression-free survival and
0
time-to-progression results were observed for mRECIST 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42
and RECIST 1.1 based on masked independent imaging Number at risk
Time (months)
review. Masked independent imaging review confirmed a Lenvatinib 478 436 374 297 253 207 178 140 102 67 40 21 8 2 0
significantly higher objective response rate in the Sorafenib 476 440 348 282 230 192 156 116 83 57 33 16 8 4 0

lenvatinib arm than in the sorafenib arm by mRECIST
and RECIST 1.1 (table 2).
156 (33%) patients in the lenvatinib arm and
184 (39%) in the sorafenib arm received post-study anti­
cancer medication (including investigational therapy). Of
these patients, 121 (25%) in the lenvatinib arm and
56 (12%) in the sorafenib arm received sorafenib during
survival follow-up. In the western region, 41 (26%) patients
in the lenvatinib arm received anticancer medication
during survival follow-up versus 61 (39%) patients in the
Figure 2: Overall survival outcomes
Kaplan-Meier estimates of overall survival by treatment group. HR=hazard ratio.
sorafenib arm. In the lenvatinib arm, 11 (7%) patients in
the western region had an anticancer procedure during
follow-up compared with 18 (11%) patients in the sorafenib
arm in this region (appendix).
The median duration of study treatment for patients in
the lenvatinib group was 5·7 months (IQR 2·9–11·1),
compared with 3·7 months (1·8–7·4) in the sorafenib

www.thelancet.com Vol 391 March 24, 2018 1167

 Articles

Lenvatinib (n=478) Sorafenib (n=476) Effect size (95% CI) p value

Investigator review according to mRECIST
Overall survival (months) 13·6 (12·1–14·9) 12·3 (10·4–13·9) HR 0·92 (0·79–1·06) ··
Progression-free survival (months) 7·4 (6·9–8·8) 3·7 (3·6–4·6) HR 0·66 (0·57−0·77) <0·0001
Time to progression (months) 8·9 (7·4–9·2) 3·7 (3·6–5·4) HR 0·63 (0·53–0·73) <0·0001
Objective response (%, 95% CI) 115 (24·1%, 20·2–27·9) 44 (9·2%, 6·6–11·8) OR 3·13 (2·15–4·56) <0·0001
Complete response 6 (1%) 2 (<1%) ·· ··
Partial response 109 (23%) 42 (9%) ·· ··
Stable disease 246 (51%) 244 (51%) ·· ··
Durable stable disease lasting ≥23 weeks 167 (35%) 139 (29%) ·· ··
Progressive disease 71 (15%) 147 (31%) ·· ··
Unknown or not evaluable 46 (10%) 41 (9%) ·· ··
Disease control rate (%, 95% CI) 361 (75·5%, 71·7–79·4) 288 (60·5%, 56·1–64·9) ·· ··
Masked independent imaging review according to mRECIST
Progression-free survival (months) 7·3 (5·6–7·5) 3·6 (3·6–3·7) HR 0·64 (0·55–0·75) <0·0001
Time to progression (months) 7·4 (7·2–9·1) 3·7 (3·6–3·9) HR 0·60 (0·51–0·71) <0·0001
Objective response (%, 95% CI) 194 (40·6%, 36·2–45·0) 59 (12·4%, 9·4–15·4) OR 5·01 (3·59–7·01) <0·0001
Complete response 10 (2%) 4 (1%) ·· ··
Partial response 184 (38%) 55 (12%) ·· ··
Stable disease 159 (33%) 219 (46%) ·· ··
Durable stable disease lasting ≥23 weeks 84 (18%) 90 (19%) ·· ··
Progressive disease 79 (17%) 152 (32%) ·· ··
Unknown or not evaluable 46 (10%) 46 (10%) ·· ··
Disease control rate (%, 95% CI) 353 (73·8%, 69·9–77·8) 278 (58·4%, 54·0–62·8) ·· ··
Masked independent imaging review according to RECIST 1.1
Progression-free survival (months) 7·3 (5·6–7·5) 3·6 (3·6–3·9) HR 0·65 (0·56–0·77) <0·0001
Time to progression (months) 7·4 (7·3–9·1) 3·7 (3·6–5·4) HR 0·61 (0·51–0·72) <0·0001
Objective response (%, 95% CI) 90 (18·8%, 15·3–22·3) 31 (6·5%, 4·3–8·7) OR 3·34 (2·17–5·14) <0·0001
Complete response 2 (<1%) 1 (<1%) ·· ··
Partial response 88 (18%) 30 (6%) ·· ··
Stable disease 258 (54%) 250 (53%) ·· ··
Durable stable disease lasting ≥23 weeks 163 (34%) 118 (25%) ·· ··
Progressive disease 84 (18%) 152 (32%) ·· ··
Unknown or not evaluable 46 (10%) 43 (9%) ·· ··
Disease control rate (%, 95% CI) 348 (72·8%, 68·8–76·8) 281 (59·0%, 54·6–63·5) ·· ··

Data are presented as median (95% CI) or n (%) unless otherwise indicated. mRECIST=modified Response Evaluation Criteria in Solid Tumours. HR=hazard ratio. OR=odds ratio.

Table 2: Efficacy measures

group. Treatment-emergent adverse events occurred in
most patients who received lenvatinib or sorafenib
(table 3). Adjusted by patient-years, the adverse event rate
was 18·9 episodes per patient-year in the lenvatinib group
and 19·7 episodes per patient-year in the sorafenib group.
Treatment-emergent adverse events of grade 3 or higher
occurred at similar rates in the lenvatinib and sorafenib
arms (episodes per patient-year 3·2 vs 3·3). The most
common treatment-emergent adverse events among
patients who received lenvatinib were hypertension,
diarrhoea, decreased appetite, and decreased weight. In
the sorafenib arm, the most common treatment-emergent
adverse events were palmar-plantar erythrodysaesthesia,
diarrhoea, hyper­tension, and decreased appetite (table 3).
Fatal adverse events occurred throughout treatment
and appeared to occur at similar rates in both arms. Fatal
adverse events determined by the investigator to be
related to lenvatinib treatment occurred in 11 (2%) patients
and included hepatic failure (three patients), cerebral
haemorrhage (three patients), and respiratory failure
(two patients). In the sorafenib group, treatment-related
fatal adverse events occurred in four (1%) patients
and included tumour haemorrhage, ischaemic stroke,
respiratory failure, and sudden death (one each).
Treatment-related treatment-emergent adverse events
led to lenvatinib drug interruption in 190 (40%) patients,
dose reduction in 176 (37%) patients, and drug withdrawal
in 42 (9%) patients. In the sorafenib arm, treatment-
related treatment-emergent adverse events led to drug
interruption in 153 (32%) patients, dose reduction in
181 (38%), and drug withdrawal in 34 (7%) patients. The
mean lenvatinib dose intensity was 7·0 mg in the

1168 www.thelancet.com Vol 391 March 24, 2018

 Articles

A Events/patients HR (95% CI) Median (months)

lenvatinib vs sorafenib
Lenvatinib Sorafenib Lenvatinib Sorafenib
Age (years)
<65 203/270 204/283 0·94 (0·77–1·15) 12·4 11·4
≥65 148/208 146/193 0·84 (0·66–1·07) 14·6 13·4
Sex
Male 293/405 293/401 0·91 (0·77–1·07) 13·4 12·4
Female 58/73 57/75 0·84 (0·56–1·26) 15·3 11·4
Region
Asia-Pacific 243/321 248/319 0·86 (0·72–1·02) 13·5 11·0
Western 108/157 102/157 1·08 (0·82–1·42) 13·6 14·2
ECOG-PS
PS=0 221/304 223/301 0·88 (0·73–1·06) 14·6 12·8
PS=1 130/174 127/175 0·97 (0·76–1·25) 10·7 10·3
Bodyweight (kg)
<60 110/153 113/146 0·85 (0·65–1·11) 13·4 10·3
≥60 241/325 237/330 0·95 (0·79–1·14) 13·7 12·5
Macroscopic portal vein invasion, extrahepatic spread, or both
Yes 250/329 259/336 0·87 (0·73–1·04) 11·5 9·8
No 101/149 91/140 1·05 (0·79–1·40) 18·0 18·0
AFP at baseline (ng/mL)
<200 167/255 193/286 0·91 (0·74–1·12) 19·5 16·3
≥200 183/222 154/187 0·78 (0·63–0·98) 10·4 8·2
Aetiology
HBV 196/259 186/244 0·83 (0·68–1·02) 13·4 10·2
HCV 75/103 97/135 0·91 (0·66–1·26) 15·3 14·1
Alcohol 22/33 15/23 1·03 (0·47–2·28) 14·1 11·9
BCLC staging
Stage B 71/104 65/92 0·91 (0·65–1·28) 18·5 17·3
Stage C 280/374 285/384 0·92 (0·77–1·08) 11·8 10·3
Post-treatment anticancer therapy
Yes 143/206 175/243 0·84 (0·67–1·06) 19·5 17·0
No 208/272 175/233 0·91 (0·74–1·11) 10·5 7·9
Post-treatment anticancer procedures
Yes 63/99 82/112 0·71 (0·51–1·01) 23·0 19·6
No 288/379 268/364 0·94 (0·79–1·11) 11·6 10·1
Post-treatment anticancer medication
Yes 110/156 132/184 0·87 (0·67–1·14) 20·8 17·0
No 241/322 218/292 0·90 (0·75–1·09) 11·5 9·1
Overall 351/478 350/476 0·92 (0·79–1·06) 13·6 12·3

0·1 1·0 10·0

Favours lenvatinib Favours sorafenib

B Events/patients HR (95% CI) Median (months)

lenvatinib vs sorafenib
Lenvatinib Sorafenib Lenvatinib Sorafenib
Age (years)
<65 201/270 223/283 0·67 (0·55–0·82) 7·3 3·6
≥65 148/208 144/193 0·61 (0·48–0·78) 7·4 5·4
Sex
Male 298/405 308/401 0·66 (0·56–0·77) 7·4 3·7
Female 51/73 59/75 0·75 (0·49–1·13) 7·4 4·6
Region
Asia-Pacific 249/321 264/319 0·61 (0·51–0·73) 7·3 3·6
Western 100/157 103/157 0·81 (0·61–1·08) 7·4 5·5
ECOG-PS
PS=0 220/304 233/301 0·63 (0·52–0·76) 7·4 3·7
PS=1 129/174 134/175 0·70 (0·55–0·90) 7·3 3·7
Bodyweight (kg)
<60 111/153 121/146 0·61 (0·46–0·79) 7·4 3·6
≥60 238/325 246/330 0·69 (0·58–0·83) 7·4 3·7
Macroscopic portal vein invasion, extrahepatic spread, or both
Yes 246/329 265/336 0·64 (0·54–0·77) 7·3 3·6
No 103/149 102/140 0·73 (0·55–0·97) 9·2 5·6
AFP at baseline (ng/mL)
<200 186/255 209/286 0·68 (0·55–0·83) 9·0 5·4
≥200 163/222 157/187 0·59 (0·47–0·75) 5·5 2·4
Aetiology
HBV 205/259 199/244 0·62 (0·50–0·75) 7·3 3·6
HCV 70/103 103/135 0·78 (0·56–1·09) 7·4 5·3
Alcohol 19/33 18/23 0·27 (0·11–0·66) 8·8 3·9
BCLC staging
Figure 3: Forest plots of
Stage B 72/104 66/92 0·70 (0·50–0·99) 9·1 5·5 overall and progression-free
Stage C 277/374 301/384 0·63 (0·53–0·75) 7·3 3·7 survival in patient subgroups
Post-treatment anticancer therapy
Yes 177/206 204/243 0·58 (0·47–0·72) 7·2 3·6 Subgroup analyses for overall
No 172/272 163/233 0·70 (0·56–0·87) 8·0 3·7 survival (A) and
Post-treatment anticancer procedures progression-free survival (B).
Yes 80/99 93/112 0·41 (0·29–0·57) 7·4 3·6
No 269/379 274/364 0·71 (0·59–0·84) 7·4 3·7 HR=hazard ratio.
Post-treatment anticancer medication ECOG-PS=Eastern Cooperative
Yes 137/156 157/184 0·66 (0·51–0·85) 5·7 3·8 Oncology Group performance
No 212/322 210/292 0·66 (0·54–0·80) 8·6 3·7
Overall 349/478 367/476 0·66 (0·57–0·77) 7·4 3·7 status. AFP=α-fetoprotein.
HBV=hepatitis B virus.
0·1 1·0 10·0 HCV=hepatitis C virus.
BCLC=Barcelona Clinic Liver
Favours lenvatinib Favours sorafenib
Cancer.

www.thelancet.com Vol 391 March 24, 2018 1169

 Articles

100 Median progression-free survival Lenvatinib Sorafenib

duration (months; 95% CI) (n=476) (n=475)
90
Lenvatinib 7·4 (6·9–8·8) (Continued from previous column)
80 Sorafenib 3·7 (3·6–4·6)
Alopecia
HR 0·66 (95% CI 0·57–0·77)
Progression-free survival (%)

70 Any grade 14 (3%) 119 (25%)

Log-rank p<0·0001
Grade ≥3 0 0
60
Proteinuria
50
Any grade 117 (25%) 54 (11%)
40 Grade ≥3 27 (6%) 8 (2%)
Dysphonia
30
Any grade 113 (24%) 57 (12%)
20
Grade ≥3 1 (<1%) 0
10 Nausea
Any grade 93 (20%) 68 (14%)
0
0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 Grade ≥3 4 (1%) 4 (1%)
Time (months) Abdominal pain
Number at risk
Lenvatinib 478 345 223 172 106 69 44 28 14 9 4 2 0 0 0 Any grade 81 (17%) 87 (18%)
Sorafenib 476 262 140 94 56 41 33 22 14 9 4 2 2 0 0
Grade ≥3 8 (2%) 13 (3%)
Figure 4: Progression-free survival outcomes Decreased platelet count
Kaplan-Meier estimates of progression-free survival by modified Response Evaluation Criteria in Solid Tumours. Any grade 87 (18%) 58 (12%)
HR=hazard ratio. Grade ≥3 26 (5%) 16 (3%)
Elevated aspartate aminotransferase
Any grade 65 (14%) 80 (17%)
Lenvatinib Sorafenib
Grade ≥3 24 (5%) 38 (8%)
(n=476) (n=475)
Hypothyroidism
Total treatment-emergent adverse events 470 (99%) 472 (99%)
Any grade 78 (16%) 8 (2%)
Total treatment-related 447 (94%) 452 (95%)
Grade ≥3 0 0
treatment-emergent adverse events
Vomiting
Treatment-emergent adverse events of 357 (75%) 316 (67%)
grade ≥3 Any grade 77 (16%) 36 (8%)
Treatment-related treatment-emergent 270 (57%) 231 (49%) Grade ≥3 6 (1%) 5 (1%)
adverse events of grade ≥3 Constipation
Serious treatment-emergent adverse events 205 (43%) 144 (30%) Any grade 76 (16%) 52 (11%)
Serious treatment-related 84 (18%) 48 (10%) Grade ≥3 3 (1%) 0
treatment-emergent adverse events
Rash
Treatment-emergent adverse events occurring in ≥15% of patients in either
Any grade 46 (10%) 76 (16%)
treatment group
Grade ≥3 0 2 (<1%)
Palmar-plantar erythrodysaesthesia
Increased blood bilirubin
Any grade 128 (27%) 249 (52%)
Any grade 71 (15%) 63 (13%)
Grade ≥3 14 (3%) 54 (11%)
Grade ≥3 31 (7%) 23 (5%)
Diarrhoea
Any grade 184 (39%) 220 (46%) Data are presented as n (%).
Grade ≥3 20 (4%) 20 (4%)
Table 3: Adverse events
Hypertension
Any grade 201 (42%) 144 (30%)
Grade ≥3 111 (23%) 68 (14%) 8 mg/day group and 10·5 mg in the 12 mg/day group,
Decreased appetite corresponding to 88% of the planned starting dose in
Any grade 162 (34%) 127 (27%) both cases. The mean sorafenib dose intensity was
Grade ≥3 22 (5%) 6 (1%) 663·8 mg, or 83% of the planned starting dose.
Decreased weight Baseline scores on the EORTC QLQ-C30 and EORTC
Any grade 147 (31%) 106 (22%) QLQ-HCC18 health questionnaires were similar in the
Grade ≥3 36 (8%) 14 (3%) lenvatinib and sorafenib treatment groups. Following
Fatigue treatment, scores declined in both groups. Analysis of time
Any grade 141 (30%) 119 (25%) to clinically meaningful deterioration showed that role
Grade ≥3 18 (4%) 17 (4%) functioning (nominal p=0·0193), pain (nominal p=0·0105),
(Table 3 continues in next column) and diarrhoea (nominal p<0·0001) from EORTC QLQ-C30,
and nutrition (nominal p=0·0113) and body image

1170 www.thelancet.com Vol 391 March 24, 2018


(nominal p=0·0051) from EORTC QLQ-HCC18 were
observed earlier in patients treated with sorafenib than
in those treated with lenvatinib. For between-group
comparison, the summary score was not significantly
different between the treatment arms (HR 0·87, 95% CI
0·754–1·013, appendix).
Based on individual model-derived predicted lenvatinib
area under the curve (AUC) values at steady state for
patients with hepatocellular carcinoma in our study, the
median values and ranges of AUC between the group
with a starting dose of 8 mg for bodyweight less than
60 kg (median 1820·2 ng·h/mL, range 704·8–4980·7)
and the group with a 12 mg starting dose for bodyweight
of at least 60 kg (1996·0 ng·h/mL, 925·5–5427·9) are
comparable, which supports a starting dose of 8 mg for
bodyweights less than 60 kg, and confirms the weight-
based dosing reported in pharmacokinetic analyses
from a previous study.19 There were no differences in
lenvatinib oral clearance or in AUC at steady state among
Western, Asian, Chinese, and Japanese populations in
our study.
Discussion
To our knowledge, our study is the first global phase 3
trial in 10 years to show a treatment effect on overall
survival, and the first ever positive trial against an active
control. Our study showed lenvatinib to be non-inferior to
sorafenib—the current standard of care in hepato­cellular
carcinoma—for overall survival. Lenvatinib showed
statistically significant clinically meaningful improvement
for all secondary efficacy endpoints (progression-free
survival, time to progression, and objective response)
across subgroups, and in quality-of-life assessments.
Together, these data support the overall survival result
of our study.
The median overall survival time of patients who
received sorafenib in our study is longer than that reported
in any previous large randomised phase 3 study.4–9
A possible explanation for this result is the high proportion
of post-sorafenib anticancer therapy in our study. For
example, in a previous phase 3 study7 of brivanib versus
sorafenib, 21% of patients who received sorafenib under­
went systemic post-sorafenib treatments and 17% had
non-systemic post-sorafenib treatments, compared with
39% of patients receiving systemic post-sorafenib
treatments and 27% of patients receiving non-systemic
post-sorafenib treatments in our study. Continuous
improvements in care for unresectable hepatocellular
carcinoma have been made, and multi­modality therapies,
including locoregional treatment approaches, are often
used after disease progression because they might be
efficacious, even after systemic therapies such as sorafenib
treatment.27,28 If post-progression survival is prolonged by
such post-study treatments, this could lead to dilution of
the observed overall survival treatment benefit. Hence,
although still representing the gold standard, overall
survival as an endpoint alone for trials in first-line
www.thelancet.com Vol 391 March 24, 2018 1171
Articles
hepatocellular carcinoma treatment might no longer
capture the full extent of antitumour efficacy. The
substantial improvement in progression-free survival,
time to progression, and objective response with lenvatinib
in our study might indicate, as in some other tumours, the
emergence of a broader framework in drug assessment
and treatment in advanced hepatocellular carcinoma.
Our study did not enrol patients with more than
50% liver involvement and main portal vein invasion
because this exclusion criterion was used in the preceding
phase 2 proof-of-concept study in Japan, as mandated by
Japan Society of Hepatology consensus-based clinical
practice guidelines.17,29 This decision resulted in only
4·2% screen failures in the phase 3 study. Although this
exclusion criterion could have slightly changed the overall
prognosis of the patient population, it did not affect the
distribution of patients between the study arms because
this was controlled for by the randomisation.
The safety profile of lenvatinib was consistent with that
observed in previous studies.16,18,30 Patients who received
lenvatinib experienced fewer instances of palmar-plantar
erythrodysaesthesia, diarrhoea, and alopecia, and more
instances of hypertension, proteinuria, dysphonia, and
hypothyroidism than did patients who received sorafenib.
Although quality-of-life scores declined in both groups
after treatment, a clinically meaningful delay in deterior­
ation for multiple domains was observed with lenvatinib
compared with sorafenib.
The median duration of lenvatinib treatment was
1·5 times longer than that of sorafenib treatment,
which might have contributed to the higher incidence
of adverse events. When adjusted for treatment
duration, almost all adverse event episodes were
comparable for the lenvatinib and sorafenib arms.
Doses of lenvatinib for hepatocellular carcinoma are
lower than the dosage for radioiodine-refractory
differentiated thyroid cancer (24 mg/day). In a phase
1 study of lenvatinib in hepatocellular carcinoma,31
patients with hepatocellular carcinoma who received
12 mg of lenvatinib per day and patients with solid
tumours who received 25 mg of lenvatinib per day had
similar lenvatinib plasma concentrations at 24 h,
possibly because lenvatinib is metabolised in the liver.
In our study, similar clinical activities and safety
profiles were observed for both the 8 mg/day and
12 mg/day lenvatinib starting doses.
Unlike other cancer types, including differentiated
thyroid cancer and renal cell carcinoma, lenvatinib
pharmacokinetics were affected by bodyweight to a
clinically significant degree. The final pharmacokinetic
model for lenvatinib included bodyweight effect as an
allometric constant on both clearance and volume
parameters, whereby both parameters increased with
increasing bodyweight. The clinical relevance of this
finding is that, when administered equivalent doses,
patients with hepatocellular carcinoma with low
bodyweight will have clinically significantly higher
 Articles
exposures than will patients with high bodyweight,
supporting bodyweight-based dosing.
Our study was potentially limited by its open-label
design. However, because of the distinct toxicities
and dose management requirements, this design
was essential to ensure patient safety. Major protocol
deviations were few and balanced, the percentage of
patients having clinical progression and drug dis­
continuations were similar in both arms, and results
were confirmed by masked independent imaging
review. Therefore, we believe any bias introduced by the
open-label design was minimal. The full analysis set
was used as the primary analysis set as opposed to the
per-protocol set. However, the sample size calculation
for our study was such that any factor introducing bias
toward the null hypothesis would reduce the power of
the study. Therefore, use of the full analysis set as the
primary analysis set for non-inferiority testing is a
conservative approach, and, in fact, overall survival
analysis based on the per-protocol set was completely
consistent with that based on the full analysis set.
Use of mRECIST could also be considered as a
limitation of this study. However, mRECIST is an
established tool in hepatocellular carcinoma.32,33 Further­
more, exploratory post-hoc analysis confirmed that
progression-free survival and time to progression based
on investigator assessment using mRECIST were similar
to those observed based on independent imaging review
using both mRECIST and RECIST 1.1.
In conclusion, this study showed non-inferiority of
lenvatinib versus sorafenib in terms of overall survival,
as well as statistically significant and clinically meaning­
ful improvement in progression-free survival, time to
progression, and objective response rate. The safety
profiles of lenvatinib and sorafenib in our study appear
consistent with the known safety profiles of these drugs
in hepatocellular carcinoma, and no new safety signals
were identified. Based on our results, lenvatinib might
be a potential new treatment option for advanced
hepatocellular carcinoma.
Contributors
MK, RSF, SQ, K-HH, KI, FP, and A-LC were protocol steering
committee members and made substantial contributions in all aspects
of ICMJE criteria. Equal contributions were made by AB, J-WP, and GH
(non-protocol steering committee member investigators). AB and J-WP
contributed to helpful communications in study management and
acquisition of good quality data, and GH contributed to substantial good
quality data acquisition and critical data interpretation of the Chinese
patient population. JJ, JFB, AV, DK, TRJE, and CL were national
coordinating or representing investigators in European countries, and
particularly contributed to study coordination and acquisition of good
quality data. CD, MG, KS, SK, TT, and MR are Eisai employees primarily
involved in the study, and played a significant role in study design, data
collection, data analysis, data interpretation, and writing of the report.
MK, RSF, SQ, K-HH, KI, FP, CD, MG, KS, TT, and A-LC contributed to
the study design. MG, KS, and MR did the statistical analysis.
Declaration of interests
MK reports honoraria from Bayer, Eisai, MSD, and EA Pharma.
RSF reports grants, personal fees and non-financial support from Eisai,
Bayer, Pfizer, Novartis, Bristol Myers Squibb (BMS), and Merck outside
1172 www.thelancet.com Vol 391 March 24, 2018
the submitted work. K-HH reports grants and consultant fees from Eisai
and KOWA, and consultant fees from Bayer, all outside the submitted
work. KI reports honoraria from Eisai and Dainippon Sumitomo
Pharma. FP reports personal fees from Eisai during the conduct of the
study, and grants and personal fees from Bayer, and personal fees from
Bracco, both outside the submitted work. AB reports research funding
from Eisai. JJ reports personal fees from AstraZeneca, Roche, Pfizer,
G1 Therapeutics, Pierre Fabre, Celgene, Merck, and BMS outside the
submitted work. JFB reports personal fees from Bayer SP, Eli Lilly
Oncology, Novartis, and BMS outside the submitted work. TRJE reports
other fees (reimbursement of study costs of this clinical trial [to the
institution]; advisory board honorarium [payable to the institution]) from
Eisai during the conduct of the study, and other fees from BMS
(financial support for clinical trials of novel anti-cancer drugs, honoraria
for consultancies or speaker’s fees, and support to attend international
conferences), Clovis (support for clinical trials [to institution] and
honorarium for advisory board), Karus Therapeutics (scientific advisory
board [payable to the institution]), Baxalta (advisory board honorarium
[payable to the institution]), Bayer (support for clinical trials and advisory
board honorarium [payable to the institution]), Celgene (support for
clinical trials and advisory board honorarium [payable to the institution]),
GlaxoSmithKline (support for clinical trials and advisory board
honorarium [payable to the institution]), Otsuka (support for clinical
trials and advisory board honorarium [payable to the institution]),
Roche/Genentech (support for clinical trials and advisory board
honorarium [payable to the institution]), TC Biopharm (support for
clinical trials), Immunova (advisory board honorarium [payable to the
institution]), Basilea (support for clinical trials), e-Therapeutics (support
for clinical trials), Immunocore (support for clinical trials), Vertex
(support for clinical trials), Verastem (support for clinical trials), Daiichi
(support for clinical trials), and Merck (support for clinical trials) outside
the submitted work. CL reports grants, personal fees, non-financial
support and advisory board fees from Eisai, Bayer, Lilly, and
Daiichi Sankyo during the conduct of the study. CD, MG, KS, SK, TT,
and MR are employees of Eisai. A-LC reports personal fees from BMS,
Ono, Novartis, Bayer, Merck, and MSD during the conduct of the study.
SQ, J-WP, GH, AV, and DK declare no competing interests.
Acknowledgments
We thank the patients, their families, the investigators, and the teams
who participated in this trial, and Terri Binder (Eisai Inc., Woodcliff
Lake, NJ, USA) for overseeing the independent image review. Editorial
assistance was provided by Nicolette Belletier of Oxford PharmaGenesis
(funded by Eisai Inc).
References
1 Ferlay J, Soerjomataram I, Dikshit R, et al. Cancer incidence and
mortality worldwide: sources, methods and major patterns in
GLOBOCAN 2012. Int J Cancer 2015; 136: E359–86.
2 El-Serag HB, Rudolph KL. Hepatocellular carcinoma: epidemiology
and molecular carcinogenesis. Gastroenterology 2007; 132: 2557–76.
3 Balogh J, Victor D 3rd, Asham EH, et al. Hepatocellular carcinoma:
a review. J Hepatocell Carcinoma 2016; 3: 41–53.
4 Llovet JM, Ricci S, Mazzaferro V, et al. Sorafenib in advanced
hepatocellular carcinoma. N Engl J Med 2008; 359: 378–90.
5 Cheng AL, Kang YK, Chen Z, et al. Efficacy and safety of sorafenib
in patients in the Asia-Pacific region with advanced hepatocellular
carcinoma: a phase III randomised, double-blind,
placebo-controlled trial. Lancet Oncol 2009; 10: 25–34.
6 Cheng AL, Kang YK, Lin DY, et al. Sunitinib versus sorafenib in
advanced hepatocellular cancer: results of a randomized
phase III trial. J Clin Oncol 2013; 31: 4067–75.
7 Johnson PJ, Qin S, Park JW, et al. Brivanib versus sorafenib as
first-line therapy in patients with unresectable, advanced
hepatocellular carcinoma: results from the randomized
phase III BRISK-FL study. J Clin Oncol 2013; 31: 3517–24.
8 Cainap C, Qin S, Huang WT, et al. Linifanib versus sorafenib in
patients with advanced hepatocellular carcinoma: results of a
randomized phase III trial. J Clin Oncol 2015; 33: 172–79.
9 Zhu AX, Rosmorduc O, Evans TR, et al. SEARCH: a phase III,
randomized, double-blind, placebo-controlled trial of sorafenib plus
erlotinib in patients with advanced hepatocellular carcinoma.
J Clin Oncol 2015; 33: 559–66.

10 Bruix J, Qin S, Merle P, et al. Regorafenib for patients with
hepatocellular carcinoma who progressed on sorafenib treatment
(RESORCE): a randomised, double-blind, placebo-controlled,
phase 3 trial. Lancet 2017; 389: 55–66.
11 NCCN. Clinical practice guidelines in oncology, version 1:
hepatobiliary cancers. Fort Washington, PA: National
Comprehensive Cancer Network, 2017.
12 Matsui J, Yamamoto Y, Funahashi Y, et al. E7080, a novel inhibitor
that targets multiple kinases, has potent antitumor activities against
stem cell factor producing human small cell lung cancer H146,
based on angiogenesis inhibition. Int J Cancer 2008; 122: 664–71.
13 Matsui J, Funahashi Y, Uenaka T, Watanabe T, Tsuruoka A, Asada M.
Multi-kinase inhibitor E7080 suppresses lymph node and lung
metastases of human mammary breast tumor MDA-MB-231 via
inhibition of vascular endothelial growth factor-receptor (VEGF-R)
2 and VEGF-R3 kinase. Clin Cancer Res 2008; 14: 5459–65.
14 Tohyama O, Matsui J, Kodama K, et al. Antitumor activity of
lenvatinib (e7080): an angiogenesis inhibitor that targets multiple
receptor tyrosine kinases in preclinical human thyroid cancer
models. J Thyroid Res 2014; 2014: 638747.
15 Yamamoto Y, Matsui J, Matsushima T, et al. Lenvatinib, an
angiogenesis inhibitor targeting VEGFR/FGFR, shows broad
antitumor activity in human tumor xenograft models associated
with microvessel density and pericyte coverage. Vasc Cell 2014; 6: 18.
16 Schlumberger M, Tahara M, Wirth LJ, et al. Lenvatinib versus
placebo in radioiodine-refractory thyroid cancer. N Engl J Med 2015;
372: 621–30.
17 Motzer RJ, Hutson TE, Glen H, et al. Lenvatinib, everolimus, and
the combination in patients with metastatic renal cell carcinoma:
a randomised, phase 2, open-label, multicentre trial. Lancet Oncol
2015; 16: 1473–82.
18 Ikeda K, Kudo M, Kawazoe S, et al. Phase 2 study of lenvatinib in
patients with advanced hepatocellular carcinoma.
J Gastroenterol 2017; 52: 512–19.
19 Tamai T, Hayato S, Hojo S, et al. Dose finding of lenvatinib in
subjects with advanced hepatocellular carcinoma based on
population pharmacokinetic and exposure-response analyses.
J Clin Pharmacol 2017; 57: 1138–47.
20 Lencioni R, Llovet JM. Modified RECIST (mRECIST) assessment
for hepatocellular carcinoma. Semin Liver Dis 2010; 30: 52–60.
21 Bruix J, Sherman M. Management of hepatocellular carcinoma:
an update. Hepatology 2011; 53: 1020–22.
22 Eisenhauer EA, Therasse P, Bogaerts J, et al. New response
evaluation criteria in solid tumours: revised RECIST guideline
(version 1.1). Eur J Cancer 2009; 45: 228–47.
www.thelancet.com Vol 391 March 24, 2018 1173
Articles
23 Cocks K, King MT, Velikova G, Martyn St-James M, Fayers PM,
Brown JM. Evidence-based guidelines for determination of sample
size and interpretation of the European Organisation for the
Research and Treatment of Cancer Quality of Life Questionnaire
Core 30. J Clin Oncol 2011; 29: 89–96.
24 Giesinger JM, Kieffer JM, Fayers PM, et al. Replication and
validation of higher order models demonstrated that a summary
score for the EORTC QLQ-C30 is robust. J Clin Epidemiol 2016;
69: 79–88.
25 Chie WC, Blazeby JM, Hsiao CF, et al. International cross-cultural
field validation of an European Organization for Research and
Treatment of Cancer questionnaire module for patients with
primary liver cancer, the European Organization for Research and
Treatment of Cancer quality-of-life questionnaire HCC18.
Hepatology 2012; 55: 1122–29.
26 National Cancer Institute. Protocol development. Cancer therapy
evaluation program. https://ctep.cancer.gov/protocolDevelopment/
electronic_applications/ctc.htm (accessed March 21, 2017).
27 Terashima T, Yamashita T, Arai K, et al. Feasibility and efficacy of
hepatic arterial infusion chemotherapy for advanced hepatocellular
carcinoma after sorafenib. Hepatol Res 2014; 44: 1179–85.
28 Shao YY, Liang PC, Wu YM, et al. A pilot study of hepatic arterial
infusion of chemotherapy for patients with advanced hepatocellular
carcinoma who have failed anti-angiogenic therapy. Liver Int 2013;
33: 1413–19.
29 Kudo M, Matsui O, Izumi N, et al. JSH consensus-based clinical
practice guidelines for the management of hepatocellular
carcinoma: 2014 update by the Liver Cancer Study Group of Japan.
Liver Cancer 2014; 3: 458–68.
30 Boss DS, Glen H, Beijnen JH, et al. A phase I study of E7080,
a multitargeted tyrosine kinase inhibitor, in patients with advanced
solid tumours. Br J Cancer 2012; 106: 1598–604.
31 Ikeda M, Okusaka T, Mitsunaga S, et al. Safety and
pharmacokinetics of lenvatinib in patients with advanced
hepatocellular carcinoma. Clin Cancer Res 2016; 22: 1385–94.
32 Lencioni R, Montal R, Torres F, et al. Objective response by
mRECIST as a predictor and potential surrogate end-point of
overall survival in advanced HCC. J Hepatol 2017; 66: 1166–72.

33 Meyer T, Palmer DH, Cheng AL, Hocke J, Loembé AB, Yen CJ.
mRECIST to predict survival in advanced hepatocellular
carcinoma: analysis of two randomised phase II trials comparing
nintedanib vs sorafenib. Liver Int 2017; 37: 1047–55.